A Comparison of Algorithms for Retrieving Soil Moisture from ENVISAT/ASAR Images

Abstract

In this paper, we present an intercomparison of algorithms for retrieving soil moisture content (SMC) from ENVIronmental SATtellite (ENVISAT)/Advanced Synthetic Aperture Radar images. The algorithms taken into consideration were a feedforward artificial neural network (ANN) with two hidden layers, a statistical approach based on Bayes' theorem, and an iterative algorithm based on the Nelder-Mead direct-search method. The comparison was carried out by using both simulated and experimental data. Simulated data were obtained by means of the Integral Equation Model (IEM). Experimental data were collected in an agricultural area in Northern Italy during 2003-2005; they included backscattering coefficient at HH and HV polarizations and at an incidence angle of θ = 23 • , as well as detailed ground truth measurements of SMC, surface roughness, and vegetation parameters. HH-polarized data were related to SMC, whereas the information of the cross-polarized channel was used to correct the backscatter for the effects of surface roughness. A comparison of the algorithms with experimental data showed that all the tested approaches produced SMC values that are very close to the measured ones. However, the predictions of the ANN were slightly more suitable than the other methods for generating maps in reasonable time. The production of moisture maps carried out at different dates using this algorithm pointed out the feasibility of separating up to six levels of spatial/temporal variations of SMC in the range of 10%-35%. Index Terms-Artificial neural network (ANN), backscattering coefficient, ENVIronmental SATtellite (ENVISAT)/AdvancedSynthetic Aperture Radar (ASAR), inversion algorithms, soil moisture content (SMC) maps.

Statistical Modeling of Soil Moisture, Integrating Satellite Remote-Sensing (SAR) and Ground-Based Data

“…There are a wide variety of existing models that can be used to predict soil moisture and integrate satellite, RS imagery data-from simpler deterministic and semi-empirical models to probabilistic optimization methods (e.g., feed-forward neural networks (ANNs), Bayesian, Nelder-Mead gradient-based approaches) [15,16]. Theoretical radiation-transfer models, such as the small perturbation model (SPM), the physical optics model (PO) and the geometrical optics model (GO) predict the radar backscatter in response to changes in surface roughness or surface (< 5 cm) soil moisture [17].…”

Soil Moisture Retrieval from Active Spaceborne Microwave Observations: An Evaluation of Current Techniques

“…In order to invert the IEM and directly relate σº to the model predictions over both bare and sparsely vegetated surfaces, several algorithms have been devised based on the fitting of IEM numerical simulations for a variety of soil moisture and roughness conditions, including Look Up Tables (LUTs) [79][80][81], Neural Networks (NN) [75,[82][83][84], Bayesian approaches [85][86][87] and minimisation techniques (Nelder-Mead minimisation method devised by Nelder and Mead [88,89] and later adapted by Paloscia et al [89]). Santi et al [90] compared the performances of three of these approaches (Bayes, Neural Networks and Nelder-Mead minimisation) and found them to yield satisfactory results, although the Nelder-Mead minimisation tended to slightly overestimate soil moisture values.…”

Evaluation of a Global Soil Moisture Product from Finer Spatial Resolution SAR Data and Ground Measurements at Irish Sites

“…In [26][27][28][29][30][31][32][33][34][35], the capability of ASAR backscattering to estimate the surface soil moisture has been demonstrated.…”